In a conventional panoramic auto-focusing algorithm, an image is provided by selecting a focal plane of a specific interest region for trajectory scan lengths with a frequency analysis and by linking the trajectories. In this case, the reconstruction is performed by selecting one trajectory for each region if depths of the focal planes of upper region/lower region (for example, the upper jaw/the lower jaw, the teeth/the roots of teeth)are different. Thus, there is a disadvantage that focal points of images cannot be perfectly reconstructed. And, an error is occurred in finding exact focal depth because an error occurrence possibility is high due to noise characteristics in an image reconstruction with the frequency analysis.